Softfloat support files for PowerPC.
This commit is contained in:
simonb
parent
339a9c8de2
commit
3f04a27329
|
|
@ -0,0 +1,48 @@
|
|||
/* $NetBSD: milieu.h,v 1.1 2001/05/25 11:38:57 simonb Exp $ */
|
||||
|
||||
/*
|
||||
===============================================================================
|
||||
|
||||
This C header file is part of the SoftFloat IEC/IEEE Floating-point
|
||||
Arithmetic Package, Release 2a.
|
||||
|
||||
Written by John R. Hauser. This work was made possible in part by the
|
||||
International Computer Science Institute, located at Suite 600, 1947 Center
|
||||
Street, Berkeley, California 94704. Funding was partially provided by the
|
||||
National Science Foundation under grant MIP-9311980. The original version
|
||||
of this code was written as part of a project to build a fixed-point vector
|
||||
processor in collaboration with the University of California at Berkeley,
|
||||
overseen by Profs. Nelson Morgan and John Wawrzynek. More information
|
||||
is available through the Web page `http://HTTP.CS.Berkeley.EDU/~jhauser/
|
||||
arithmetic/SoftFloat.html'.
|
||||
|
||||
THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE. Although reasonable effort
|
||||
has been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT
|
||||
TIMES RESULT IN INCORRECT BEHAVIOR. USE OF THIS SOFTWARE IS RESTRICTED TO
|
||||
PERSONS AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ANY
|
||||
AND ALL LOSSES, COSTS, OR OTHER PROBLEMS ARISING FROM ITS USE.
|
||||
|
||||
Derivative works are acceptable, even for commercial purposes, so long as
|
||||
(1) they include prominent notice that the work is derivative, and (2) they
|
||||
include prominent notice akin to these four paragraphs for those parts of
|
||||
this code that are retained.
|
||||
|
||||
===============================================================================
|
||||
*/
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Include common integer types and flags.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
#include "powerpc-gcc.h"
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Symbolic Boolean literals.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
enum {
|
||||
FALSE = 0,
|
||||
TRUE = 1
|
||||
};
|
||||
|
|
@ -0,0 +1,89 @@
|
|||
/* $NetBSD: powerpc-gcc.h,v 1.1 2001/05/25 11:38:57 simonb Exp $ */
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
One of the macros `BIGENDIAN' or `LITTLEENDIAN' must be defined.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
#include <machine/endian.h>
|
||||
#if _BYTE_ORDER == _BIG_ENDIAN
|
||||
#define BIGENDIAN
|
||||
#endif
|
||||
#if _BYTE_ORDER == _LITTLE_ENDIAN
|
||||
#define LITTLEENDIAN
|
||||
#endif
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
The macro `BITS64' can be defined to indicate that 64-bit integer types are
|
||||
supported by the compiler.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
#define BITS64
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Each of the following `typedef's defines the most convenient type that holds
|
||||
integers of at least as many bits as specified. For example, `uint8' should
|
||||
be the most convenient type that can hold unsigned integers of as many as
|
||||
8 bits. The `flag' type must be able to hold either a 0 or 1. For most
|
||||
implementations of C, `flag', `uint8', and `int8' should all be `typedef'ed
|
||||
to the same as `int'.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
typedef int flag;
|
||||
typedef int uint8;
|
||||
typedef int int8;
|
||||
typedef int uint16;
|
||||
typedef int int16;
|
||||
typedef unsigned int uint32;
|
||||
typedef signed int int32;
|
||||
#ifdef BITS64
|
||||
typedef unsigned long long int uint64;
|
||||
typedef signed long long int int64;
|
||||
#endif
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Each of the following `typedef's defines a type that holds integers
|
||||
of _exactly_ the number of bits specified. For instance, for most
|
||||
implementation of C, `bits16' and `sbits16' should be `typedef'ed to
|
||||
`unsigned short int' and `signed short int' (or `short int'), respectively.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
typedef unsigned char bits8;
|
||||
typedef signed char sbits8;
|
||||
typedef unsigned short int bits16;
|
||||
typedef signed short int sbits16;
|
||||
typedef unsigned int bits32;
|
||||
typedef signed int sbits32;
|
||||
#ifdef BITS64
|
||||
typedef unsigned long long int bits64;
|
||||
typedef signed long long int sbits64;
|
||||
#endif
|
||||
|
||||
#ifdef BITS64
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
The `LIT64' macro takes as its argument a textual integer literal and
|
||||
if necessary ``marks'' the literal as having a 64-bit integer type.
|
||||
For example, the GNU C Compiler (`gcc') requires that 64-bit literals be
|
||||
appended with the letters `LL' standing for `long long', which is `gcc's
|
||||
name for the 64-bit integer type. Some compilers may allow `LIT64' to be
|
||||
defined as the identity macro: `#define LIT64( a ) a'.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
#define LIT64( a ) a##LL
|
||||
#endif
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
The macro `INLINE' can be used before functions that should be inlined. If
|
||||
a compiler does not support explicit inlining, this macro should be defined
|
||||
to be `static'.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
#define INLINE static __inline
|
||||
|
||||
#define FLOAT64_DEMANGLE(a) (a)
|
||||
#define FLOAT64_MANGLE(a) (a)
|
||||
|
|
@ -0,0 +1,473 @@
|
|||
|
||||
/*
|
||||
===============================================================================
|
||||
|
||||
This C source fragment is part of the SoftFloat IEC/IEEE Floating-point
|
||||
Arithmetic Package, Release 2a.
|
||||
|
||||
Written by John R. Hauser. This work was made possible in part by the
|
||||
International Computer Science Institute, located at Suite 600, 1947 Center
|
||||
Street, Berkeley, California 94704. Funding was partially provided by the
|
||||
National Science Foundation under grant MIP-9311980. The original version
|
||||
of this code was written as part of a project to build a fixed-point vector
|
||||
processor in collaboration with the University of California at Berkeley,
|
||||
overseen by Profs. Nelson Morgan and John Wawrzynek. More information
|
||||
is available through the Web page `http://HTTP.CS.Berkeley.EDU/~jhauser/
|
||||
arithmetic/SoftFloat.html'.
|
||||
|
||||
THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE. Although reasonable effort
|
||||
has been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT
|
||||
TIMES RESULT IN INCORRECT BEHAVIOR. USE OF THIS SOFTWARE IS RESTRICTED TO
|
||||
PERSONS AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ANY
|
||||
AND ALL LOSSES, COSTS, OR OTHER PROBLEMS ARISING FROM ITS USE.
|
||||
|
||||
Derivative works are acceptable, even for commercial purposes, so long as
|
||||
(1) they include prominent notice that the work is derivative, and (2) they
|
||||
include prominent notice akin to these four paragraphs for those parts of
|
||||
this code that are retained.
|
||||
|
||||
===============================================================================
|
||||
*/
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Underflow tininess-detection mode, statically initialized to default value.
|
||||
(The declaration in `softfloat.h' must match the `int8' type here.)
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
#ifdef SOFTFLOAT_FOR_GCC
|
||||
static
|
||||
#endif
|
||||
int8 float_detect_tininess = float_tininess_after_rounding;
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Raises the exceptions specified by `flags'. Floating-point traps can be
|
||||
defined here if desired. It is currently not possible for such a trap to
|
||||
substitute a result value. If traps are not implemented, this routine
|
||||
should be simply `float_exception_flags |= flags;'.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
void float_raise( int8 flags )
|
||||
{
|
||||
|
||||
float_exception_flags |= flags;
|
||||
|
||||
}
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Internal canonical NaN format.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
typedef struct {
|
||||
flag sign;
|
||||
bits64 high, low;
|
||||
} commonNaNT;
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
The pattern for a default generated single-precision NaN.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
#define float32_default_nan 0xFFFFFFFF
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Returns 1 if the single-precision floating-point value `a' is a NaN;
|
||||
otherwise returns 0.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
#ifdef SOFTFLOAT_FOR_GCC
|
||||
static
|
||||
#endif
|
||||
flag float32_is_nan( float32 a )
|
||||
{
|
||||
|
||||
return ( 0xFF000000 < (bits32) ( a<<1 ) );
|
||||
|
||||
}
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Returns 1 if the single-precision floating-point value `a' is a signaling
|
||||
NaN; otherwise returns 0.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
flag float32_is_signaling_nan( float32 a )
|
||||
{
|
||||
|
||||
return ( ( ( a>>22 ) & 0x1FF ) == 0x1FE ) && ( a & 0x003FFFFF );
|
||||
|
||||
}
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Returns the result of converting the single-precision floating-point NaN
|
||||
`a' to the canonical NaN format. If `a' is a signaling NaN, the invalid
|
||||
exception is raised.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
static commonNaNT float32ToCommonNaN( float32 a )
|
||||
{
|
||||
commonNaNT z;
|
||||
|
||||
if ( float32_is_signaling_nan( a ) ) float_raise( float_flag_invalid );
|
||||
z.sign = a>>31;
|
||||
z.low = 0;
|
||||
z.high = ( (bits64) a )<<41;
|
||||
return z;
|
||||
|
||||
}
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Returns the result of converting the canonical NaN `a' to the single-
|
||||
precision floating-point format.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
static float32 commonNaNToFloat32( commonNaNT a )
|
||||
{
|
||||
|
||||
return ( ( (bits32) a.sign )<<31 ) | 0x7FC00000 | ( a.high>>41 );
|
||||
|
||||
}
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Takes two single-precision floating-point values `a' and `b', one of which
|
||||
is a NaN, and returns the appropriate NaN result. If either `a' or `b' is a
|
||||
signaling NaN, the invalid exception is raised.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
static float32 propagateFloat32NaN( float32 a, float32 b )
|
||||
{
|
||||
flag aIsNaN, aIsSignalingNaN, bIsNaN, bIsSignalingNaN;
|
||||
|
||||
aIsNaN = float32_is_nan( a );
|
||||
aIsSignalingNaN = float32_is_signaling_nan( a );
|
||||
bIsNaN = float32_is_nan( b );
|
||||
bIsSignalingNaN = float32_is_signaling_nan( b );
|
||||
a |= 0x00400000;
|
||||
b |= 0x00400000;
|
||||
if ( aIsSignalingNaN | bIsSignalingNaN ) float_raise( float_flag_invalid );
|
||||
if ( aIsNaN ) {
|
||||
return ( aIsSignalingNaN & bIsNaN ) ? b : a;
|
||||
}
|
||||
else {
|
||||
return b;
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
The pattern for a default generated double-precision NaN.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
#define float64_default_nan LIT64( 0xFFFFFFFFFFFFFFFF )
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Returns 1 if the double-precision floating-point value `a' is a NaN;
|
||||
otherwise returns 0.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
#ifdef SOFTFLOAT_FOR_GCC
|
||||
static
|
||||
#endif
|
||||
flag float64_is_nan( float64 a )
|
||||
{
|
||||
|
||||
return ( LIT64( 0xFFE0000000000000 ) <
|
||||
(bits64) ( FLOAT64_DEMANGLE(a)<<1 ) );
|
||||
|
||||
}
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Returns 1 if the double-precision floating-point value `a' is a signaling
|
||||
NaN; otherwise returns 0.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
flag float64_is_signaling_nan( float64 a )
|
||||
{
|
||||
|
||||
return
|
||||
( ( ( FLOAT64_DEMANGLE(a)>>51 ) & 0xFFF ) == 0xFFE )
|
||||
&& ( FLOAT64_DEMANGLE(a) & LIT64( 0x0007FFFFFFFFFFFF ) );
|
||||
|
||||
}
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Returns the result of converting the double-precision floating-point NaN
|
||||
`a' to the canonical NaN format. If `a' is a signaling NaN, the invalid
|
||||
exception is raised.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
static commonNaNT float64ToCommonNaN( float64 a )
|
||||
{
|
||||
commonNaNT z;
|
||||
|
||||
if ( float64_is_signaling_nan( a ) ) float_raise( float_flag_invalid );
|
||||
z.sign = FLOAT64_DEMANGLE(a)>>63;
|
||||
z.low = 0;
|
||||
z.high = FLOAT64_DEMANGLE(a)<<12;
|
||||
return z;
|
||||
|
||||
}
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Returns the result of converting the canonical NaN `a' to the double-
|
||||
precision floating-point format.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
static float64 commonNaNToFloat64( commonNaNT a )
|
||||
{
|
||||
|
||||
return FLOAT64_MANGLE(
|
||||
( ( (bits64) a.sign )<<63 )
|
||||
| LIT64( 0x7FF8000000000000 )
|
||||
| ( a.high>>12 ) );
|
||||
|
||||
}
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Takes two double-precision floating-point values `a' and `b', one of which
|
||||
is a NaN, and returns the appropriate NaN result. If either `a' or `b' is a
|
||||
signaling NaN, the invalid exception is raised.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
static float64 propagateFloat64NaN( float64 a, float64 b )
|
||||
{
|
||||
flag aIsNaN, aIsSignalingNaN, bIsNaN, bIsSignalingNaN;
|
||||
|
||||
aIsNaN = float64_is_nan( a );
|
||||
aIsSignalingNaN = float64_is_signaling_nan( a );
|
||||
bIsNaN = float64_is_nan( b );
|
||||
bIsSignalingNaN = float64_is_signaling_nan( b );
|
||||
a |= FLOAT64_MANGLE(LIT64( 0x0008000000000000 ));
|
||||
b |= FLOAT64_MANGLE(LIT64( 0x0008000000000000 ));
|
||||
if ( aIsSignalingNaN | bIsSignalingNaN ) float_raise( float_flag_invalid );
|
||||
if ( aIsNaN ) {
|
||||
return ( aIsSignalingNaN & bIsNaN ) ? b : a;
|
||||
}
|
||||
else {
|
||||
return b;
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
#ifdef FLOATX80
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
The pattern for a default generated extended double-precision NaN. The
|
||||
`high' and `low' values hold the most- and least-significant bits,
|
||||
respectively.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
#define floatx80_default_nan_high 0xFFFF
|
||||
#define floatx80_default_nan_low LIT64( 0xFFFFFFFFFFFFFFFF )
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Returns 1 if the extended double-precision floating-point value `a' is a
|
||||
NaN; otherwise returns 0.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
flag floatx80_is_nan( floatx80 a )
|
||||
{
|
||||
|
||||
return ( ( a.high & 0x7FFF ) == 0x7FFF ) && (bits64) ( a.low<<1 );
|
||||
|
||||
}
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Returns 1 if the extended double-precision floating-point value `a' is a
|
||||
signaling NaN; otherwise returns 0.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
flag floatx80_is_signaling_nan( floatx80 a )
|
||||
{
|
||||
bits64 aLow;
|
||||
|
||||
aLow = a.low & ~ LIT64( 0x4000000000000000 );
|
||||
return
|
||||
( ( a.high & 0x7FFF ) == 0x7FFF )
|
||||
&& (bits64) ( aLow<<1 )
|
||||
&& ( a.low == aLow );
|
||||
|
||||
}
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Returns the result of converting the extended double-precision floating-
|
||||
point NaN `a' to the canonical NaN format. If `a' is a signaling NaN, the
|
||||
invalid exception is raised.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
static commonNaNT floatx80ToCommonNaN( floatx80 a )
|
||||
{
|
||||
commonNaNT z;
|
||||
|
||||
if ( floatx80_is_signaling_nan( a ) ) float_raise( float_flag_invalid );
|
||||
z.sign = a.high>>15;
|
||||
z.low = 0;
|
||||
z.high = a.low<<1;
|
||||
return z;
|
||||
|
||||
}
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Returns the result of converting the canonical NaN `a' to the extended
|
||||
double-precision floating-point format.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
static floatx80 commonNaNToFloatx80( commonNaNT a )
|
||||
{
|
||||
floatx80 z;
|
||||
|
||||
z.low = LIT64( 0xC000000000000000 ) | ( a.high>>1 );
|
||||
z.high = ( ( (bits16) a.sign )<<15 ) | 0x7FFF;
|
||||
return z;
|
||||
|
||||
}
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Takes two extended double-precision floating-point values `a' and `b', one
|
||||
of which is a NaN, and returns the appropriate NaN result. If either `a' or
|
||||
`b' is a signaling NaN, the invalid exception is raised.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
static floatx80 propagateFloatx80NaN( floatx80 a, floatx80 b )
|
||||
{
|
||||
flag aIsNaN, aIsSignalingNaN, bIsNaN, bIsSignalingNaN;
|
||||
|
||||
aIsNaN = floatx80_is_nan( a );
|
||||
aIsSignalingNaN = floatx80_is_signaling_nan( a );
|
||||
bIsNaN = floatx80_is_nan( b );
|
||||
bIsSignalingNaN = floatx80_is_signaling_nan( b );
|
||||
a.low |= LIT64( 0xC000000000000000 );
|
||||
b.low |= LIT64( 0xC000000000000000 );
|
||||
if ( aIsSignalingNaN | bIsSignalingNaN ) float_raise( float_flag_invalid );
|
||||
if ( aIsNaN ) {
|
||||
return ( aIsSignalingNaN & bIsNaN ) ? b : a;
|
||||
}
|
||||
else {
|
||||
return b;
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
#endif
|
||||
|
||||
#ifdef FLOAT128
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
The pattern for a default generated quadruple-precision NaN. The `high' and
|
||||
`low' values hold the most- and least-significant bits, respectively.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
#define float128_default_nan_high LIT64( 0xFFFFFFFFFFFFFFFF )
|
||||
#define float128_default_nan_low LIT64( 0xFFFFFFFFFFFFFFFF )
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Returns 1 if the quadruple-precision floating-point value `a' is a NaN;
|
||||
otherwise returns 0.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
flag float128_is_nan( float128 a )
|
||||
{
|
||||
|
||||
return
|
||||
( LIT64( 0xFFFE000000000000 ) <= (bits64) ( a.high<<1 ) )
|
||||
&& ( a.low || ( a.high & LIT64( 0x0000FFFFFFFFFFFF ) ) );
|
||||
|
||||
}
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Returns 1 if the quadruple-precision floating-point value `a' is a
|
||||
signaling NaN; otherwise returns 0.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
flag float128_is_signaling_nan( float128 a )
|
||||
{
|
||||
|
||||
return
|
||||
( ( ( a.high>>47 ) & 0xFFFF ) == 0xFFFE )
|
||||
&& ( a.low || ( a.high & LIT64( 0x00007FFFFFFFFFFF ) ) );
|
||||
|
||||
}
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Returns the result of converting the quadruple-precision floating-point NaN
|
||||
`a' to the canonical NaN format. If `a' is a signaling NaN, the invalid
|
||||
exception is raised.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
static commonNaNT float128ToCommonNaN( float128 a )
|
||||
{
|
||||
commonNaNT z;
|
||||
|
||||
if ( float128_is_signaling_nan( a ) ) float_raise( float_flag_invalid );
|
||||
z.sign = a.high>>63;
|
||||
shortShift128Left( a.high, a.low, 16, &z.high, &z.low );
|
||||
return z;
|
||||
|
||||
}
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Returns the result of converting the canonical NaN `a' to the quadruple-
|
||||
precision floating-point format.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
static float128 commonNaNToFloat128( commonNaNT a )
|
||||
{
|
||||
float128 z;
|
||||
|
||||
shift128Right( a.high, a.low, 16, &z.high, &z.low );
|
||||
z.high |= ( ( (bits64) a.sign )<<63 ) | LIT64( 0x7FFF800000000000 );
|
||||
return z;
|
||||
|
||||
}
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Takes two quadruple-precision floating-point values `a' and `b', one of
|
||||
which is a NaN, and returns the appropriate NaN result. If either `a' or
|
||||
`b' is a signaling NaN, the invalid exception is raised.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
static float128 propagateFloat128NaN( float128 a, float128 b )
|
||||
{
|
||||
flag aIsNaN, aIsSignalingNaN, bIsNaN, bIsSignalingNaN;
|
||||
|
||||
aIsNaN = float128_is_nan( a );
|
||||
aIsSignalingNaN = float128_is_signaling_nan( a );
|
||||
bIsNaN = float128_is_nan( b );
|
||||
bIsSignalingNaN = float128_is_signaling_nan( b );
|
||||
a.high |= LIT64( 0x0000800000000000 );
|
||||
b.high |= LIT64( 0x0000800000000000 );
|
||||
if ( aIsSignalingNaN | bIsSignalingNaN ) float_raise( float_flag_invalid );
|
||||
if ( aIsNaN ) {
|
||||
return ( aIsSignalingNaN & bIsNaN ) ? b : a;
|
||||
}
|
||||
else {
|
||||
return b;
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
#endif
|
||||
|
|
@ -0,0 +1,293 @@
|
|||
|
||||
/*
|
||||
===============================================================================
|
||||
|
||||
This C header file is part of the SoftFloat IEC/IEEE Floating-point
|
||||
Arithmetic Package, Release 2a.
|
||||
|
||||
Written by John R. Hauser. This work was made possible in part by the
|
||||
International Computer Science Institute, located at Suite 600, 1947 Center
|
||||
Street, Berkeley, California 94704. Funding was partially provided by the
|
||||
National Science Foundation under grant MIP-9311980. The original version
|
||||
of this code was written as part of a project to build a fixed-point vector
|
||||
processor in collaboration with the University of California at Berkeley,
|
||||
overseen by Profs. Nelson Morgan and John Wawrzynek. More information
|
||||
is available through the Web page `http://HTTP.CS.Berkeley.EDU/~jhauser/
|
||||
arithmetic/SoftFloat.html'.
|
||||
|
||||
THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE. Although reasonable effort
|
||||
has been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT
|
||||
TIMES RESULT IN INCORRECT BEHAVIOR. USE OF THIS SOFTWARE IS RESTRICTED TO
|
||||
PERSONS AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ANY
|
||||
AND ALL LOSSES, COSTS, OR OTHER PROBLEMS ARISING FROM ITS USE.
|
||||
|
||||
Derivative works are acceptable, even for commercial purposes, so long as
|
||||
(1) they include prominent notice that the work is derivative, and (2) they
|
||||
include prominent notice akin to these four paragraphs for those parts of
|
||||
this code that are retained.
|
||||
|
||||
===============================================================================
|
||||
*/
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
The macro `FLOATX80' must be defined to enable the extended double-precision
|
||||
floating-point format `floatx80'. If this macro is not defined, the
|
||||
`floatx80' type will not be defined, and none of the functions that either
|
||||
input or output the `floatx80' type will be defined. The same applies to
|
||||
the `FLOAT128' macro and the quadruple-precision format `float128'.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
/* #define FLOATX80 */
|
||||
/* #define FLOAT128 */
|
||||
|
||||
#include <machine/ieeefp.h>
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Software IEC/IEEE floating-point types.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
typedef unsigned int float32;
|
||||
typedef unsigned long long float64;
|
||||
#ifdef FLOATX80
|
||||
typedef struct {
|
||||
unsigned short high;
|
||||
unsigned long long low;
|
||||
} floatx80;
|
||||
#endif
|
||||
#ifdef FLOAT128
|
||||
typedef struct {
|
||||
unsigned long long high, low;
|
||||
} float128;
|
||||
#endif
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Software IEC/IEEE floating-point underflow tininess-detection mode.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
extern int8 float_detect_tininess;
|
||||
enum {
|
||||
float_tininess_after_rounding = 0,
|
||||
float_tininess_before_rounding = 1
|
||||
};
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Software IEC/IEEE floating-point rounding mode.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
extern int8 float_rounding_mode;
|
||||
enum {
|
||||
float_round_nearest_even = FP_RN,
|
||||
float_round_to_zero = FP_RZ,
|
||||
float_round_down = FP_RM,
|
||||
float_round_up = FP_RP
|
||||
};
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Software IEC/IEEE floating-point exception flags.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
extern int8 float_exception_flags;
|
||||
enum {
|
||||
float_flag_inexact = FP_X_IMP,
|
||||
float_flag_underflow = FP_X_UFL,
|
||||
float_flag_overflow = FP_X_OFL,
|
||||
float_flag_divbyzero = FP_X_DZ,
|
||||
float_flag_invalid = FP_X_INV
|
||||
};
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Routine to raise any or all of the software IEC/IEEE floating-point
|
||||
exception flags.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
void float_raise( int8 );
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Software IEC/IEEE integer-to-floating-point conversion routines.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
float32 int32_to_float32( int );
|
||||
float64 int32_to_float64( int );
|
||||
#ifdef FLOATX80
|
||||
floatx80 int32_to_floatx80( int );
|
||||
#endif
|
||||
#ifdef FLOAT128
|
||||
float128 int32_to_float128( int );
|
||||
#endif
|
||||
float32 int64_to_float32( long long );
|
||||
float64 int64_to_float64( long long );
|
||||
#ifdef FLOATX80
|
||||
floatx80 int64_to_floatx80( long long );
|
||||
#endif
|
||||
#ifdef FLOAT128
|
||||
float128 int64_to_float128( long long );
|
||||
#endif
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Software IEC/IEEE single-precision conversion routines.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
int float32_to_int32( float32 );
|
||||
int float32_to_int32_round_to_zero( float32 );
|
||||
unsigned int float32_to_uint32_round_to_zero( float32 );
|
||||
long long float32_to_int64( float32 );
|
||||
long long float32_to_int64_round_to_zero( float32 );
|
||||
float64 float32_to_float64( float32 );
|
||||
#ifdef FLOATX80
|
||||
floatx80 float32_to_floatx80( float32 );
|
||||
#endif
|
||||
#ifdef FLOAT128
|
||||
float128 float32_to_float128( float32 );
|
||||
#endif
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Software IEC/IEEE single-precision operations.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
float32 float32_round_to_int( float32 );
|
||||
float32 float32_add( float32, float32 );
|
||||
float32 float32_sub( float32, float32 );
|
||||
float32 float32_mul( float32, float32 );
|
||||
float32 float32_div( float32, float32 );
|
||||
float32 float32_rem( float32, float32 );
|
||||
float32 float32_sqrt( float32 );
|
||||
flag float32_eq( float32, float32 );
|
||||
flag float32_le( float32, float32 );
|
||||
flag float32_lt( float32, float32 );
|
||||
flag float32_eq_signaling( float32, float32 );
|
||||
flag float32_le_quiet( float32, float32 );
|
||||
flag float32_lt_quiet( float32, float32 );
|
||||
flag float32_is_signaling_nan( float32 );
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Software IEC/IEEE double-precision conversion routines.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
int float64_to_int32( float64 );
|
||||
int float64_to_int32_round_to_zero( float64 );
|
||||
unsigned int float64_to_uint32_round_to_zero( float64 );
|
||||
long long float64_to_int64( float64 );
|
||||
long long float64_to_int64_round_to_zero( float64 );
|
||||
float32 float64_to_float32( float64 );
|
||||
#ifdef FLOATX80
|
||||
floatx80 float64_to_floatx80( float64 );
|
||||
#endif
|
||||
#ifdef FLOAT128
|
||||
float128 float64_to_float128( float64 );
|
||||
#endif
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Software IEC/IEEE double-precision operations.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
float64 float64_round_to_int( float64 );
|
||||
float64 float64_add( float64, float64 );
|
||||
float64 float64_sub( float64, float64 );
|
||||
float64 float64_mul( float64, float64 );
|
||||
float64 float64_div( float64, float64 );
|
||||
float64 float64_rem( float64, float64 );
|
||||
float64 float64_sqrt( float64 );
|
||||
flag float64_eq( float64, float64 );
|
||||
flag float64_le( float64, float64 );
|
||||
flag float64_lt( float64, float64 );
|
||||
flag float64_eq_signaling( float64, float64 );
|
||||
flag float64_le_quiet( float64, float64 );
|
||||
flag float64_lt_quiet( float64, float64 );
|
||||
flag float64_is_signaling_nan( float64 );
|
||||
|
||||
#ifdef FLOATX80
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Software IEC/IEEE extended double-precision conversion routines.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
int floatx80_to_int32( floatx80 );
|
||||
int floatx80_to_int32_round_to_zero( floatx80 );
|
||||
long long floatx80_to_int64( floatx80 );
|
||||
long long floatx80_to_int64_round_to_zero( floatx80 );
|
||||
float32 floatx80_to_float32( floatx80 );
|
||||
float64 floatx80_to_float64( floatx80 );
|
||||
#ifdef FLOAT128
|
||||
float128 floatx80_to_float128( floatx80 );
|
||||
#endif
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Software IEC/IEEE extended double-precision rounding precision. Valid
|
||||
values are 32, 64, and 80.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
extern int floatx80_rounding_precision;
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Software IEC/IEEE extended double-precision operations.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
floatx80 floatx80_round_to_int( floatx80 );
|
||||
floatx80 floatx80_add( floatx80, floatx80 );
|
||||
floatx80 floatx80_sub( floatx80, floatx80 );
|
||||
floatx80 floatx80_mul( floatx80, floatx80 );
|
||||
floatx80 floatx80_div( floatx80, floatx80 );
|
||||
floatx80 floatx80_rem( floatx80, floatx80 );
|
||||
floatx80 floatx80_sqrt( floatx80 );
|
||||
flag floatx80_eq( floatx80, floatx80 );
|
||||
flag floatx80_le( floatx80, floatx80 );
|
||||
flag floatx80_lt( floatx80, floatx80 );
|
||||
flag floatx80_eq_signaling( floatx80, floatx80 );
|
||||
flag floatx80_le_quiet( floatx80, floatx80 );
|
||||
flag floatx80_lt_quiet( floatx80, floatx80 );
|
||||
flag floatx80_is_signaling_nan( floatx80 );
|
||||
|
||||
#endif
|
||||
|
||||
#ifdef FLOAT128
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Software IEC/IEEE quadruple-precision conversion routines.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
int float128_to_int32( float128 );
|
||||
int float128_to_int32_round_to_zero( float128 );
|
||||
long long float128_to_int64( float128 );
|
||||
long long float128_to_int64_round_to_zero( float128 );
|
||||
float32 float128_to_float32( float128 );
|
||||
float64 float128_to_float64( float128 );
|
||||
#ifdef FLOATX80
|
||||
floatx80 float128_to_floatx80( float128 );
|
||||
#endif
|
||||
|
||||
/*
|
||||
-------------------------------------------------------------------------------
|
||||
Software IEC/IEEE quadruple-precision operations.
|
||||
-------------------------------------------------------------------------------
|
||||
*/
|
||||
float128 float128_round_to_int( float128 );
|
||||
float128 float128_add( float128, float128 );
|
||||
float128 float128_sub( float128, float128 );
|
||||
float128 float128_mul( float128, float128 );
|
||||
float128 float128_div( float128, float128 );
|
||||
float128 float128_rem( float128, float128 );
|
||||
float128 float128_sqrt( float128 );
|
||||
flag float128_eq( float128, float128 );
|
||||
flag float128_le( float128, float128 );
|
||||
flag float128_lt( float128, float128 );
|
||||
flag float128_eq_signaling( float128, float128 );
|
||||
flag float128_le_quiet( float128, float128 );
|
||||
flag float128_lt_quiet( float128, float128 );
|
||||
flag float128_is_signaling_nan( float128 );
|
||||
|
||||
#endif
|
||||
Loading…
Reference in New Issue